On Tue, Sep 13, 2005 at 10:58:13AM -0700, Jason Thorpe wrote:
> The fundamental problem is that tmpfs is using two pages of each page  
> of data -- one in the vnode, one in the (swap-backed) aobj.

I agree. I was wondering if one could tell the buffer cache not buffer
pages from a certain VFS ...

> The pages should simply be always associated with the vnode, with no  
> aobj.  I think tmpfs_getpages() and tmpfs_putpages() should interact  
> with swap slots directly, cleaning the vnode pages to and faulting in  
> from swap.

... but that sounds like a much better approach.

	Kind regards

-- 
Matthias Scheler                                  http://scheler.de/~matthias/